DE3751412T2 - Method and device for image processing with gradation correction of the image signal. - Google Patents

Description

Background of the invention Field of the invention

The present invention relates to a method and an apparatus for processing an image, and more particularly, to a method and an apparatus for processing an image in which gradation is corrected in an efficient manner when a still image represented by a video signal is recorded in the form of a visualized image on an image recording medium.

Description of the state of the art

In the following, prior art image processing methods as shown in JP-A-6037878 and JP-A-5756843 are described.

There has been proposed a still image recording apparatus which receives a video signal read from a video signal recording medium such as a floppy disk or a video tape to reproduce a visual image on an image recording medium such as a photographic printing paper.

In such a device, for example, the red signals R, the green signals G and the blue signals B, each obtained from the input video signal, are sequentially supplied to a recording monochrome cathode ray tube having a high brightness. In front of a screen of the recording cathode ray tube, a lens and a color separation filter are arranged for separating three colors to focus an image displayed on the screen on a color printing sheet. Color components including cyan C, magenta M and yellow Y are developed in the color printing sheet to obtain a color image. In this case, the level between the R, G and B signals obtained from the input video signal and the signal supplied to the monochrome cathode ray tube having a high brightness must be inverted to develop the color elements C, M and Y. That is, for example, for a bright part of an image in which the R, G and B signals have a high level, it is required that the output level of the part be minimized because the development of colors by the color components C, M, Y must be reduced to represent the brightness.

Furthermore, for an image represented by an input video signal, since the density of each color differs from that of the real object depending on, for example, lighting conditions and a camera used to record the object, compensation for the difference in density is necessary. For example, even if the red, green, and blue colors each have the same intensity at a white part having the highest level in a real object, the R, G, and B signals obtained may each have different maximum levels in some cases. In such a situation, if the input R, G, and B signals are directly used, for example, a part to be printed in white may be recorded on a recording medium in a color that is different due to the differences between the levels of the R, G and B signals are similar to either red, green or blue.

To compensate for this phenomenon, gradation correction was performed using a look-up table for each of the input R, G and B signals. Namely, a look-up table containing data of output levels corresponding to the input signals is used to effect gradation correction, which further performs the level inversion operation described above.

In the current art, the lookup table data is determined as follows.

For the R, G and B video signals obtained from the input video signal, points RH, GH and BH near the highest points of the respective levels are selected to represent a highlight point, and input video signal levels RH, GH and BH at the respective points are converted to the same density DH which is a thin density similar to a density having the lowest level, thereby generating video signals at the same density.

Similarly, when points RS, GS and BS are selected in the vicinity of points having the respective lowest levels of the R, G and B video signals to represent a shadow point, the input video signal levels RS, GS and BS at these points are converted to the same density D which is a "thin" density near the density having the highest level, thereby generating video signals at the same level.

As described above, in a graph of an output density versus an input video signal level, the highlight point and the shadow point are determined for each of the R, G and B signals, a curve is selected from a plurality of curves prepared in advance, the curve having the shortest distances to the highlight points and shadow points is selected, and then the data on the selected curve is used as data for the look-up table to perform gradation correction. That is, for example, from a plurality of curves, a plurality of curves having the shortest distance to the highlight point are selected, and then a curve having the smallest distance to the shadow point is selected from the selected curves.

The prior art lookup table data obtained in the manner of selecting a curve is obtained from a limited number of curves, and therefore the data is unlikely to pass through the highlight and shadow points when graphically displayed correctly; thus, an error occurs that the lookup table data is unsuitable for gradation correction of an image represented by an input video signal.

Since the data of many curves are stored, there is also the disadvantage that storage devices with a large capacity are necessary.

The highlight and shadow points are obtained, for example, by generating a cumulative histogram that represents a frequency distribution of a video signal of each pixel in an image (frame) of an input video signal.

For example, the highlight point is set to a point where a value of 99% is displayed in the cumulative histogram. That is, 90% of the input signals of the pixels have levels that are below the level of an input video signal assigned to the highlight point.

Similarly, the shadow point is set to a point where a value of 1% is displayed in the cumulative histogram.

Consequently, in both cases where a part of the input video signals having a level near the level of the highlight point occupies a large area and a small area, the gradation is similarly determined with the output density D near the highlight point. However, since the gradation in the area near the highlight point contributes greatly to the picture quality of the entire image, if there is a large area having a level near the level of the highlight point, the large area part attracts the attention of the viewer of the captured image; therefore, it is desirable to attach much importance to the gradation in the large area part so that the level difference of the input video signals with respect to the output density is increased in the display. On the other hand, when there is a small area having a level close to the level of the highlight point, the small area part does not attract the attention of the viewer of the captured image and therefore it is desirable not to attach much importance to the gradation in the small area part so that the level difference of the input video signals with respect to the output density in the representation is reduced in order to create a hard gradation in the overall image.

In order to change the gradation in the output density depending on the area near the highlight point, the highlight point must be determined taking into account the area near the highlight point. Similarly, the area must be taken into account to set the shadow point.

However, since in the prior art the highlight point and the shadow point are respectively fixed to the 99% and 1% points in the cumulative histogram as already described above, an appropriate gradation correction taking the area into account cannot be achieved.

Furthermore, since the highlight and shadow points are set only to the 99% and 1% points in the cumulative histogram, the highlight and shadow points are under the influence of noise and contour emphasis, which prevents proper gradation correction when the number of sampling points is reduced to generate the cumulative histogram. On the other hand, in a method of referring to the entire frequency distribution in a histogram to obtain a normalized distribution in the histogram, an area with a medium density is also referred to, resulting in a strong dependence on a scene and therefore resulting in an unnatural image; as a result, correct color matching cannot be achieved in many cases.

Incidentally, in a case where an image contains a white part, the lookup table data is as follows For each of the R, G and B input signals, assuming that the point at the highest level and the point at the lowest level are the highlight and shadow points, respectively, the same output levels are obtained at the highlight and shadow points, respectively. With this measure, for the white part, the R, G and B signals are generated with the same intensity, and therefore a white surface is obtained.

However, when, for example, an object is photographed using an electronic still camera, an input video signal is generally subjected to processing to emphasize the contour. As a result, when a cumulative histogram is generated with the video signal of each pixel included in the input video signal, the cumulative histogram includes an influence from a pixel whose level has been changed due to the contour emphasis, making it impossible to correctly determine the highlight and shadow points.

Furthermore, since the input video signal is usually mixed with noise, the cumulative histogram is also under the influence of the noise, which further leads to a disadvantage in that the determination of the highlight and shadow points cannot be carried out properly.

Incidentally, there has been a method of creating a look-up table to correct the gradation of the R, G and B signals, in which the cumulative histogram is created with the luminance signal Y of the input video signal to obtain the highlight and shadow points from the histogram associated with the luminance signal Y to use these points as highlight and shadow points of the R, G and B signals. In a case of this method, since the differences between the R, G and B signals contained in the input video signal are not taken into account, the difference in color tone between the input signal and the actual object cannot be corrected.

Furthermore, there has been a method of creating a look-up table to correct the gradation of the R, G and B signals, in which a cumulative histogram is created for each of the R, G and B signals obtained from the input video signal to determine the highlight and shadow points using the created histograms. According to this method, for example, in a case where a part of the R signal has a relatively high level, namely, when a bright part appears in an image, the highlight point of the R signal is set to a higher value compared with the G and B signals and thus the intensity of the output of the R signal is lowered; as a result, the complementary color with respect to the R signal, namely, cyan, is increased in the recorded image and the color balance is impaired.

In other words, although according to this method, there is no problem in a case where an image represented by the input video signal has a pure white portion, the above-described influence caused by the input signal occurs when the image has no such white portion and an input signal having a high level for a pure color with a high chroma saturation exists. As a result, an image for which the gradation has been corrected with the look-up table has an unsatisfactory color balance. That is, the correction of the Gradation with the determined lookup table results in overcorrection and the (color) tone of the image differs from that of the real object.

EP-A-0163903 describes a density correction method for a subtraction image. Characteristic values of image signals are calculated, for example, the values of the minimum and maximum signals are calculated using a histogram of the image signals. The characteristic values are used to correct a curve stored in a look-up table.

The prior art gradation correction as described above has a disadvantage in that the characteristic values (highlight point, shadow point) are determined as fixed values (e.g., minimum value, maximum value). This may result in insufficient gradation correction for an image that has, for example, only a small area with a level near the maximum value. Consequently, it is desirable to give great consideration to the gradation in the large area part because the viewer's attention is mainly drawn to the large area part of the image. The above-mentioned conventional gradation correction does not take into account what proportion of the image area has levels near the minimum value and the maximum value.

Summary of the invention

It is an object of the present invention to provide a method and an apparatus for image processing in which optimal lookup table data is obtained in order to achieve an effective gradation correction.

This object is solved by the features of claims 1 and 15.

The present invention obtains lookup table data for gradation correction by determining at least either the highlight point or the shadow point depending on at least two points on the cumulative histogram, these two points being near the highlight point or the shadow point to be determined, and by converting the standard lookup table data by a function transformation. A preferred embodiment of the present invention is shown in Figures 2A to 2B.

The method of the present invention has the advantage that, when the gradation correction is carried out, the shooting condition of the image is taken into account. In other words, the gradation correction is carried out taking into account the portion of the image area that has levels near the highlight point or the shadow point. For example, when the highlight portion covers a large area of the image, the contrast for the large area portion can be increased. This is advantageous because the viewer's attention is attracted to the large area portion.

Short description of the drawings

The objects and features of the present invention will become apparent from the following detailed description taken in conjunction with the accompanying drawings. In the drawings:

Figures 1A to 1B are schematic block diagrams which, when combined as shown in Figure 1, show an embodiment of an image recording system using an image processing apparatus according to the present invention;

Figures 2A to 2B are schematic block diagrams which, when combined as shown in Figure 2, show an alternative embodiment of an image recording system utilizing an image processing apparatus according to the present invention;

Figures 3A to 3B are schematic block diagrams which, when combined as shown in Figure 3, show another alternative embodiment of an image recording system utilizing an image processing device according to the present invention;

Figures 4A to 4B are schematic block diagrams which, when combined as shown in Figure 4, show another alternative embodiment of an image recording system using an image processing apparatus according to the present invention;

Figure 5 is a graph showing an example of lookup tables LUTR, LUTG and LUTB stored in a gradation corrector of Figure 1;

Figure 6 is a graph showing an example of a cumulative histogram of an input video signal of Figure 1;

Figure 7 is a graph showing a standard lookup table L0 stored in a standard lookup table memory of Figure 1;

Figure 8 is a graph showing an example of a cumulative histogram of an input video signal of Figure 1; and

Figure 9 is a graph useful for comparing different levels of an input signal near a highlight point in the cumulative histogram.

Description of the preferred embodiments

Referring now to the drawings, a description will be given of embodiments of a method and an apparatus for processing an image according to the present invention.

Figures 1A to 1B show an embodiment of an image recording system using an image processing apparatus according to the present invention.

For example, from a video signal source of a video disk playback system, video signals in the form of the three primary color signals R, G and B are supplied to the input terminals 11R, 11G and 11B of analog-to-digital, A/D converters 10R, 10G and 10B, respectively. The A/D converters 10R, 10G and 10B convert the input video signals 11R, 11G and 11B in response to a synchronizing signal SYNC 13 to supply output signals 15R, 15G and 15B, respectively, to the image memories (frame memories) 12R, 12G and 12B, respectively.

The image memories (frame memories) 12R, 12G and 12B are used to store signal data of the respective pixels, which constitute a frame of signal data of the pixels that make up an image. The output signals 17R, 17G and 17B are supplied from the image memories (frame memories) 12R, 12G and 12B to the gradation correctors 14R, 14G and 14B, respectively.

The lookup tables LUTR, LUTG and LUTB for gradation correction are provided in the gradation correctors 14R, 14G and 14B, respectively, which are parameter correctors for effecting gradation correction. The lookup tables LUTR, LUTG and LUTB are generated in the lookup table converters 34R, 34G and 34B, respectively, and are provided to the gradation correctors 14R, 14G and 14B, respectively, which will be explained later.

The look-up tables LUTR, LUTG and LUTB for gradation correction are each formed from data to be converted into signals supplied to a high-brightness cathode ray tube for causing the color components C, M and Y to develop colors in a sheet of color photographic paper 30. By conversions of signal levels using the look-up tables LUTR, LUTG and LUTB, compensation is effected for the difference due to the illumination condition used when forming an image represented by the input video signal.

Next, a description of the lookup tables LUTR, LUTG and LUTB for gradation correction is given.

Figure 5 shows data associated with examples of the lookup tables LUTR, LUTG and LUTB. In this graph, the The abscissa and ordinate respectively represent a level of an input video signal v and a level of a density signal D to be generated. By using such look-up tables LUTR, LUTG and LUTB provided for the signals R, G and B, the input video signal v is converted into an output signal representing the density D for R, G and B, respectively.

In Figure 5, the points RH, GH and BH, respectively, closest to the highest levels of the input video signals R, G and B, represent the highlight points, whereas the points RS, GS and BS, respectively, closest to the lowest levels of the R, G and B signals, indicate the shadow points. The output signals for the input video signals R, G and B, corresponding to the highlight points RH, GH and BH, have a density D set to a value DH. Similarly, the output signals for the input video signals R, G and B, corresponding to the shadow points RS, GS and BS, have a density D set to a value DS. In a case where such curves are formed as those of Figure 5 representing the data of the lookup tables LUTR, LUTG and LUTB, great importance is given to the highlight points RH, GH and BH and the shadow points RS, GS and BS.

For example, for the lookup table LUTR, a curve is drawn so that it passes through two points that contain the highlight point (RH, DH) and the shadow point (RS, DS).

The highlight and shadow points are obtained as follows.

For each of the input video signals R, G and B, a cumulative histogram is generated as shown in Figure 6. The graph in Figure 6 represents a frequency distribution of video signals from each pixel associated with a frame of an image. In this graph, the abscissa indicates a level of the input video signal, whereas the ordinate indicates, at a point on the histogram curve, a ratio (%) of the accumulated number of pixels having video signal levels below the relevant input video signal level to the total number of pixels contained in a frame of an image.

Thus, the highlight point RH in Figure 6 represents that 99% of the video signals of the pixels have signal levels lower than the level RH. The highlight point described above is usually set to a point for which 95% to 99% is indicated in the cumulative histogram. Similarly, the shadow point RS in Figure 6 represents that 1% of the video signals of the pixels have signal levels lower than the level of RS, and the shadow point is usually set to a point for which 1% to 5% is indicated in the cumulative histogram.

For the highlight and shadow points thus determined, the predetermined values DH and DS are set as the initial density, respectively. Using these density values, a curve is drawn from the look-up table LUTR so that it passes through two points of the highlight point (RH, DH) and the shadow point (RS, DS).

Back to Figure 1; the outputs 19 with 19R, 19G and 19B, for which the gradation has been corrected by the gradation corrector 14R, 14G and 14B, respectively, are each controlled by the color correctors 16R, 16G, and 16B, respectively; through the gradation correctors 18R, 18G, and 18B, respectively; and through the D/A converters 20R, 20G, and 20B, respectively, to be alternatively selected by means of a switch 22. The selected signal is supplied to the recording CRT having a high brightness 24. A lens 26 and a 3-color decomposition filter 28 are arranged in front of a display screen of the recording CRT 24 to focus an image displayed on the screen on a sheet of color photographic paper 30.

The color correctors 16R, 16G and 16B are each loaded with color correction matrices MTXS to form parameter correction sections for compensating for differences between the color tone characteristics of photosensitive materials and that of the signal source of the input video signal. In a case where, for example, a television camera is used as the input signal source, the color correction matrix contains matrix coefficients for compensating for the difference between the color tone characteristics of the television camera and that of the color photographic paper 30. With this measure, the outputs 21 contain video signal data in which the color tone has been corrected to obtain a target density.

The gradation correctors 18R, 18G and 18B respectively include gradation correction look-up tables LUT-2R, LUT-2G and LUT-28 to form parameter correction sections for compensating the gradation characteristics of the recording CRT 24 and the photographic paper 30. The outputs 23 from the gradation correctors 18R, 18G and 18B are supplied to the D/A converters 20R, 20G and 20B respectively to be converted into corresponding analog signals. These analog signals are output to the recording CRT via the switch 22. CRT 24. Switch 22 is a selection circuit which receives the three-color decomposed signals R, G and B output from D/A converters 20R, 20G and 20B, respectively, to alternatively select a signal to be supplied to recording CRT 24.

The output signals 17R, 17G and 17B from the image memories (frame memories) 12R, 12G and 12B are supplied to the cumulative histogram generators 32R, 32G and 32B, respectively. In the cumulative histogram generators 32R, 32G and 32B, signal data of all pixels constituting one frame (field) of an image are arranged according to the levels of the signals. For each input video signal, a ratio of the accumulated number of signals having levels lower than a level of an input video signal to the total number of input video signals is calculated to supply levels from the outputs 27R, 27G, and 27B to the LUT (look up table) converters 34R, 34G, and 34B, respectively, which indicate the highlight point at 99% to 95% and the shadow point at 1% to 5%, respectively. Incidentally, the signals 17R, 17G, and 17B supplied to the cumulative histogram generators 32R, 32G, and 32B, respectively, do not necessarily have to be the data of all pixels representing one frame (field) of an image, and these signals may be formed from sample data.

Furthermore, the LUT converters 34R, 34G and 34B are supplied with an output from a standard lookup table memory 36 which stores a lookup table L0 to be used as a standard of gradation correction.

The LUT converters 34R, 34G and 348 convert the output 29 of the standard LO lookup table, which is provided by the The outputs 31R, 31G and 31B of the LUT converters 34R, 34G and 34B are fed to the gradation correctors 14R, 14G and 14B, respectively, which effect gradation correction of the input video signals 17R, 17G and 17B, respectively, using the look-up table data fed from the LUT converters 34R, 34G and 34B, respectively.

Next, a description will be given of an operation for generating the lookup table data in the LUT converters 34R, 34G and 34B.

The standard lookup table data L0 to be input from the standard lookup table memory 36 to the LUT converters 34R, 34G and 34B gives a curve as shown in Figure 7. In this embodiment, the curve is expressed by

D = -2.2 log v

where v is a level of the input video signal, D indicates an output density and a coefficient value of 2.2 is a reciprocal of the gamma value for the video system.

In the LUT converters 34R, 34G and 34B, the standard curve L0 is subjected to a function transformation according to the following formula to obtain the lookup tables LUTR, LUTG and LUTB.

L(v) = S[c L0{a (v+b)} + d] ... (1)

In this expression, L0 is a standard curve of a look-up table represented by D = -2.2 log v, v denotes a level of an input video signal, a to d are transformation parameters, S [ ] indicates a transformation coefficient used to achieve a nonlinear transformation, and L(v) is an output density.

It is not necessary to use all the transformation parameters a to d, and a combination of only a and b or c and d can be used. In addition, the function transformation using the function S [ ] can be omitted. In a case where only the parameters a and b are included, c=1 and d=0 are assumed to be the expression

L(v) = L0{a(v+b)} ... (2)

to obtain.

Furthermore, if only the parameter d is used, a = 1 and b = 0 are assumed to reduce the expression to

L(v) = c L0(v) + d ... (3)

to reduce.

A case is described where, for example, using expression (2) with the transformation parameters a and b, the standard curve L0 is transformed by the function to generate the lookup table LUTR, which uses to correct the gradation of the input video signal R selected from the three primary color signals constituting the input video signal.

The following expressions are obtained by substituting the values of the highlight points (RH, DH) and the shadow points (RS, DS) contained in the output 27R from the cumulative histogram generator (32R) into the expression (2).

DH = L0{a (RH+b)} ... (21)

DS = L0{a (RS+b)} ... (22)

By substituting a and b obtained from the simultaneous system of linear equations into the expression (2), the look-up table curve LUTR is obtained. As can be seen from the expression (2), the transformation parameters a and b respectively indicate a dynamic range and a position of a black level at the rising end of the histogram.

Following the generation of the lookup table curve LUTR, the lookup table curves LUTG and LUTB are generated in a similar manner. The outputs 31R, 31G and 31B representing the lookup tables LUTR, LUTG and LUTB, respectively, obtained in the LUT converters 34R, 34G and 34B, respectively, are passed to the gradation correctors 14R, 14G and 14B, respectively.

The gradation correctors 14R, 14G and 14B use the lookup tables LUTR, LUTG and LUTB to Gradation correction on the input video signals 17R, 17G and 17B to produce the output signals 19R, 19G and 19B, respectively, which are then passed through the color correctors 16R, 16G and 16B, respectively; the gradation correctors 18R, 18G and 18B, respectively; and the D/A converters 20R, 20G and 20B, respectively, to be alternatively selected by the switch 22. The selected signal is then supplied to the recording cathode ray tube 24, which produces a color image on a sheet of photographic paper 30.

According to this embodiment, since the curves of the look-up tables used for the gradation correction of the input video signals 17R, 17G and 17B in the gradation correctors 14R, 14G and 14B are set to pass through the highlight and shadow points of the cumulative histogram, the curves of the look-up tables are correct at these points.

Furthermore, since a function transformation is performed for the standard lookup table curve LO to obtain the lookup tables without reference to a middle part of the cumulative histogram, correct gradation correction data is obtained that does not depend on the image scene even in the middle part between the brightest part and the darkest part. This prevents the captured image from becoming unnatural.

In addition, the standard LUT memory 36 only needs to be loaded with the standard lookup table curve L0, thus there is no need to store a plurality of lookup table curves as is necessary in the conventional system, and therefore the capacity of the memory can be minimized.

Figures 2A to 2B show an alternative embodiment of the image recording system using an image processing apparatus according to the present invention.

In this embodiment, the outputs 27R, 27G, and 27B from the cumulative histogram generators 32R, 32G, and 32B, respectively, are forwarded to the highlight/shadow point calculators 38R, 38G, and 38B, respectively.

The highlight/shadow point calculators 38R, 38G, and 38B, respectively, calculate the highlight and shadow points using the outputs 27R, 27G, and 27B from the cumulative histogram generators 32R, 32G, and 32B, respectively, described later, to pass the outputs 33R, 33G, and 33B8, respectively, to the LUT converters 34R, 34G, and 34B, respectively.

The LUT converters 34R, 34G and 34B convert the output 29 of a standard lookup table L0 read from a standard lookup table memory 36, taking into account the outputs 33R, 33G and 33B of the highlight and shadow points provided by the highlight/shadow point calculators 38R, 38G and 38B, to generate lookup table data for the image.

Next, a description will be given of an operation for generating the cumulative histograms in the cumulative histogram generators 32R, 32G and 32B and an operation for calculating the highlight and shadow points in the highlight/shadow point generators 38R, 38G and 38B.

In the cumulative histogram generators 32R, 32G and 32B, the cumulative histograms for the respective Input video signals R, G and B as shown in Figure 8.

The pixels used to generate the cumulative histograms need not include all the pixels that make up a frame (field) of an image, and sample pixels can also be used.

In Figure 8, the levels of the input video signals associated with the points having frequency values of 99%, 95% and 90% are represented by C1, C2 and C3, respectively.

The highlight/shadow point calculators 38R, 38G and 38B receive as inputs C1, C2 and C3 values generated by the cumulative histogram generators 32R, 32G and 32B, respectively, to calculate a highlight point CH according to the following expression.

CH = t C2 + (1-t)C1 ... (1)

where,

t = (C2-C3)/(C1-C3) ... (2).

That is, according to the expression (2), a ratio t of a level difference between the points C2 and C3 to the level difference between the points C1 and C3 is determined to substitute the resulting value for t into the expression (1), thus obtaining the highlight point CH. As can be seen from the expression (1), the highlight point CH is set to a middle point between the point C1 with a frequency of 99% and the point C2 with a frequency of 95%.

Figure 9 shows three types of cumulative histograms. The histogram A in the graph of Figure 9 has a large slope between points C3 and C2, and the slope is small between points C2 and C1. Since the ratio t between the level difference between points C2 and C3 and the level difference between points C1 and C3 in the histogram A is small, the highlight point CHC calculated by expression (1) is located at a point near point C1, as shown in Figure 9.

Since the histogram C, unlike the histogram A, has a large value for the ratio t between the level difference between the points C2 and C3 and that between the points C1 and C3, the highlight point CHC calculated by the expression (1) is set at a position near the point C2 as shown in Figure 9.

Furthermore, the histogram B lies between the histograms A and C and therefore the value of t takes an intermediate value; thus, the highlight point CHB calculated by the expression (1) lies substantially at an intermediate point between the points C1 and C2, as shown in Figure 9.

Furthermore, substitution of expression (2) into expression (1) gives

CH = C2(C2-C3)/(C1-C3) + C1(C1-C2)/(C1-C3)

As can be seen from the look-up table curves in Figure 5, when the input video signal level is above the level near the highlight point or exceeds the level of the highlight point, the ratio of the Output density difference to input video signal difference is small. That is, the contrast of output density is small; therefore, in order to increase the contrast of output density, the highlight point must be set to a point which has a high level in the input video signal.

For the histogram B in Figure 9, the highlight point is set to a value between the points C1 and C2. In a general image, in many cases, the histogram has a shape like that of the histogram B. In a case where the highlight portion occupies a large area, since the points C2 and C3 are near the point C1, the highlight point CHB also approaches the point C1, and thus the contrast of the highlight portion is greatly taken into account. On the other hand, when the highlight portion has a small area, since the points C2 and C3 are far from the point C1, the highlight point CHB is also far from the point C1, and the contrast of the highlight portion is taken into account to a small extent. As a result, the overall image is displayed in a hard tone with a satisfactory gradation.

For the histogram A, since there is a small level difference of the input video signals between the frequency of 90% and the frequency of 95%, the portion for which the input video signal is at each level has a large area. As described above, the highlight point CHA of the histogram A is set to a position near the point C1. As a result, the contrast of the output density can be increased for a large area portion near the highlight portion between the frequency of 90% and the frequency of 95%.

In the case of the histogram C, since there is a large level difference of the input video signals between the frequency of 90% and the frequency of 95%, the portion for which the input video signal is at each level has a small area. As described above, the highlight point CHC of the histogram C is located at a position near the point C2; as a result, the contrast of the output density for the portion between the frequency of 90% and the frequency of 95% is reduced. However, since this portion has a small area and therefore attracts less attention from the viewer, the reduced contrast does not cause any problems.

For the histogram C, the level difference of the input video signals between the frequency of 90% and the frequency of 95% is small, and therefore the portion of the relevant level has a large area and the highlight point CHC is formed near the point C2. As a result, the contrast in this large area part is also reduced; however, since a small level difference of the input signals occurs in this portion, no problem arises even if the highlight point CHC is located anywhere between the points C1 and C2.

Although the curves in histograms A and B of Figure 9 have a different slope between points C3 and C2 and between points C2 and C1, the ordinary curve has a substantially the same slope as histogram B, and therefore, from expression (2) it follows that t = 1/2. Consequently, expression (1) for finding the highlight point reduces to

CM = 1/2(C2+C1) ... (3)

As a result, in a case where the slope of the curve of a histogram does not have such an abrupt change as that shown at the point C2 of the histograms A and B of Figure 9, the highlight point CH can be calculated by the expression (3).

The above description of the procedure has been given with reference to an example of a highlight point. Similarly, for a shadow point, calculation of an appropriate point must be carried out using only the levels of the input video signal at frequencies of 1%, 5% and 10%. For a case of a shadow point according to the look-up table of Figure 5, although the level difference of the input signals near the shadow point is emphasized in the output density, the large level difference of the output density is minimized by subsequent processing and therefore correction for increasing the gradation is also necessary when a large area exists near the shadow point.

Assuming that the output density D associated with each of the highlight and shadow points thus determined are predetermined values DH and DS, a look-up table curve LUTR passing through the two points, for example, the highlight point (RH, DH) and the shadow point (RS, DS), is drawn based on the values DH and DS as described above.

According to the embodiment, for example, the highlight point is calculated with the expression (1) to be set at a point between the frequency of 99% and the frequency of 95% in the cumulative histogram. In the operation of setting the highlight point, referring to the input video signal levels of the frequencies of 99%, 95% and 90% in the cumulative histogram, the highlight region, namely, the size of an area of the proportion for which the input video signal levels are between 99% and 95% in the histogram, and an area near the highlight point are judged according to the value t, namely, the size of an area of the proportion for which the input video signal levels are between 95% and 90% in the histogram.

In a general image, when a region including the highlight region and a region near the highlight point, namely the region in the range of 99% to 90%, has a large area, the highlight point is set to a point corresponding to an input video signal level near 99%, and the gradation in the highlight region can be greatly taken into account. On the other hand, when this portion has a small area, the highlight point is determined at a point with a large distance from the point as 99%, and a reduced gradation can be set for the highlight region to obtain a hard tone in the overall image.

Furthermore, in a case where the region near the highlight, namely, the region in the range of 95% to 90%, has a large area, as the value of t is decreased, the highlight point CH is determined by the expression (1) to be set to a position near the point corresponding to the frequency of 99% in the cumulative histogram. Consequently, the region near the highlight, namely, the portion of the input video signals having levels between 95% and 90% in the cumulative histogram, has a large area. If it is desired to emphasize the gradation in this part, the highlight point can be set to a large value to obtain the desired gradation.

In contrast, when the part of the input video signals having levels between 95% and 90% has a small area in the cumulative histogram, as the value of t increases, the highlight point is calculated by the expression (1) to be set at a position near 95% in the cumulative histogram. Consequently, in a case where the part of the input video signals having levels near the level of the highlight point has a small area and it is desired to obtain a hard color tone without emphasizing the gradation in that part, the highlight point can be lowered to suppress the gradation.

As a result, for example, a desired gradation can be set in the part of the input video signals having levels close to the level of the highlight point according to the area of that part.

For example, if the point of 99% in the cumulative histogram is fixed as the highlight point without regard to the area of the part with the levels near the highlight point, the part has a fixed gradation. Consequently, even if the area is large and it is desired that the gradation be increased, the gradation cannot be obtained in the desired size; whereas, even if the area is small and a hard gradation is desired without emphasizing the gradation, a relatively strong gradation is obtained.

According to the embodiment as described above, an appropriate gradation can be set according to the image since the position at which the highlight point is detected varies depending on the area of the portion having levels near the level of the highlight point.

In addition, for example, in a case where the highlight point is determined only by using the point associated with the 99% frequency in the cumulative histogram, there is a fear that the highlight point is determined at a wrong position when input video signals having levels at points containing noise or representing an emphasized contour.

According to the present invention, since three points of 99%, 95% and 90% are taken into consideration when determining the highlight point, the possibility of influence of noise is reduced and therefore an appropriate highlight point having a desired gradation can be set depending on the image.

Figures 3A to 3B show another alternative embodiment of an image recording system using an image processing apparatus according to the present invention.

In this embodiment, A/D converters 10R, 10G and 10B respectively provide outputs 15R, 15G and 15B respectively to sharpening circuits 40R, 40G and 40B respectively.

The sharpness circuits 40R, 40G, and 40B receive video signals from the A/D converters 10R, 10G, and 10B, respectively, to achieve a desired sharpness or softness for the video signals. The sharpness circuits 40R, 40G, and 40B, respectively, are loaded with sharpness-emphasizing coefficients to form matrix circuits that set the sharpness for the input digital video signals 15R, 15G, and 15B, respectively.

For example, when the sharpness-emphasizing coefficients are set to emphasize the sharpness of the input signals, the contour emphasis is effected for the input video signals 15R, 15G and 15B. That is, for contour parts where signal levels vary, the signal levels are processed to change abruptly, thereby emphasizing the contour.

When the sharpening coefficients are set to weaken the sharpness of the input video signals 15R, 15G and 15B, these signals in which the contour parts have been emphasized are converted back to the original input video signals 15R, 15G and 15B before the contour emphasis. That is, the input video signals 15R, 15G and 15B are processed so that the signal levels of the contour parts thus emphasized change more slowly; and thus the emphasis of the contour is undone. The sharpening circuits 40R, 40G, 40B supply the outputs 39R, 39G and 39B to the frame memories 12R, 12G and 12B, respectively.

The D/A converters 20R, 20G and 20B deliver the outputs 25R, 25G and 25B to a switch 22 on the one hand and to a video monitor 80 on the other hand.

The outputs 25R, 25G and 25B obtained from the switch 22 are alternately selected to be supplied to a recording monochrome cathode ray tube 24 having a high brightness.

The switch 22 is a selection circuit which alternatively selects one of the three decomposed color signals 25R, 25G and 25B from the D/A converters 20R, 20G and 20V, respectively, to supply the selected signal to the recording CRT 24.

In addition, the analog signal outputs 25R, 25G, and 25B from the D/A converters 20R, 20G, and 20B, respectively, are supplied to the video monitor 80, which displays an image after gradation correction. The operator checks the displayed image and performs the necessary operations, which will be described later.

A controller 82 is a control unit for controlling the operation of the entire system and includes, for example, a processing system such as a microprocessor. The controller 82 is connected to the sharpness circuits 40R, 40G and 40B; the color correctors 16R, 16G and 16B; the gradation correctors 18R, 18G and 18B; and a memory 84 which stores values of the sharpness-emphasizing coefficients to be used in the sharpness processing and various parameter values to be set as the color correction matrix MTXA and look-up tables LUT-2R, LUT-2G and LUT-2B.

The controller 82 is further connected to an input unit 86, such as a keyboard, to be used by an operator to, for example, select or change the values for the sharpening coefficients and the image processing parameters such as the look-up tables LUT-2R and to give a necessary instruction such as an instruction to record an image.

In this embodiment, when generating a cumulative histogram, the operator sets the sharpness-enhancing coefficients in the sharpness circuits 40R, 40G and 40B to the values that determine the sharpness of the input video signals 15R, 15G and 15B. In a case where the A/D converters 10R, 10G, 10B supply the sharpening circuits 40R, 40G and 40B with input signals having an emphasized contour, the contour emphasis is removed from the video signals in the sharpening circuits 40R, 40G and 40B to return the input signals to the original signals without the contour emphasis. Furthermore, in the sharpening circuits 40R, 40G and 40B, the part where the level of the input video signals changes abruptly is processed to produce a smooth change, namely, a part having a high frequency is removed and thus the input video signals 15R, 15G and 15B are freed from noise.

As a result, since the cumulative histogram generators 32R, 32G and 32B generate a cumulative histogram with the original signals freed from the contour emphasis and noise, appropriate highlight and shadow points are obtained with the histogram. The look-up table converters 34R, 34G and 34B generate, on the basis of the highlight and shadow points thus determined, the look-up tables LUTR, LUTG and LUTB, which therefore make it possible to effect correct gradation correction of the input video signals.

After the look-up tables LUTR, LUTG and LUTB have been determined as described above, the operator changes the sharpness-emphasizing coefficients of the sharpness circuits 4R, 40G and 40B so that the sharpness circuits 40R, 40G and 40B output the input video signals 15R, 15G and 15B without weakening the sharpness of these signals. As a result, the input video signals 15R, 15G and 15B with the contour emphasis supplied by the A/D converters 10R, 10G and 10B, respectively, are output by the sharpness circuits 40R, 40G and 40B are not changed so as to be directly provided to the frame memories (image memories) 12R, 12G and 12B, respectively. If it is desired to further enhance the sharpness, the operator only needs to set the sharpness-emphasizing coefficients of the sharpness circuits 40R, 40G and 40B to enhance the contour of the input video signals.

For the input video signals 17R, 17G and 17B with the contour emphasis output from the frame memories (image memories) 12R, 12G and 12B, respectively, the gradation correctors 14R, 14G and 14B can achieve appropriate gradation correction using the look-up tables obtained as described above.

In the conventional system, such removal of the contour emphasis from the input video signals 15R, 15G and 15B, which have been subjected to contour emphasis, is not achieved by the sharpness circuits 40R, 40G and 40B. The signals with the contour emphasis are supplied to the cumulative histogram generators 32R, 32G and 32B, respectively, to generate the cumulative histograms. Furthermore, the signals containing the noise are directly supplied to the cumulative histogram generators 32R, 32G and 32B, respectively.

As a result, the highlight point or shadow point determined with the cumulative histogram is not correct due to the influence of the contour emphasis and noise, and therefore, appropriate gradation correction cannot be effected with the lookup tables LUTR, LUTG and LUTB created with the highlight or shadow point.

According to the embodiment described above, the correct highlight and shadow point can be determined, which enables the gradation correction to be properly carried out since the influence of the contour emphasis and noise are eliminated when the cumulative histograms are generated.

Furthermore, since the contour emphasis sharpening circuits 40R, 40G and 40B are also used to remove the influence of the contour emphasis from the input video signals containing the contour emphasis, it is not necessary to provide means for removing the contour emphasis and, consequently, the efficiency of the system is improved.

In a case where the contour emphasis is to be suppressed by means of the sharpness circuits 40R, 40G and 40B, signals of a small area part in the image are removed; however, since the small area part has only a small influence on a person's vision, this does not cause any problem.

Although in the above embodiment, the sharpness-emphasizing coefficients are set in the sharpness circuits 40R, 40G and 40B to remove the influence of the contour emphasis and the noise from the input video signals, a low-pass filter may be used instead of the sharpness circuits 40R, 40G and 40B. In this case, the input video signals are passed through the low-pass filter, which then removes a high frequency component to suppress the contour emphasis and the noise.

The input video signals from which the contour emphasis and noise have been removed are supplied to the cumulative histogram generators 32R, 32G and 32B to generate cumulative to generate histograms. The input video signals 17R, 17G and 17B with the emphasized contour are supplied to the gradation correctors 14R, 14G and 14B, which effect gradation correction on the input signals to obtain a print copy.

Furthermore, in a case where the input video signals contain luminance signals or color difference signals, after the processing for removing the contour emphasis is carried out only on the luminance signals in the sharpness circuits 40R, 40G and 40B or by using a low-pass filter, the color signals R, G and B can be detected from the luminance signals or the color difference signals to obtain the respective cumulative histograms. The above operation suppresses the noise in the luminance signals which may contain noise, and therefore the noise of the video signals to be supplied to the cumulative histogram generators 32R, 32G and 32B can be reduced, thus making it possible to generate the cumulative histograms in a correct manner.

Figures 4A to 4B show another alternative embodiment of an image recording system using an image processing apparatus according to the present invention.

In this embodiment, the frame memories (image memories) 12R, 12G and 12B respectively supply output signals 17R, 17G and 17B to a luminance signal generator 42 and a corrector setting device 44.

The luminance signal generator 42 generates with the output signals 17R, 17G or 17B from the frame memories (image memories) 12R, 12G and 12B respectively, a luminance signal Y according to an expression Y = 0.3R + 0.59G + 0.11B to send an output 41 to the correction setting means 44 which includes a decoding matrix for setting the amount of color correction in the output signals 17R, 17G and 17B from the frame memories 12R, 12G and 12B respectively using the luminance signal from the luminance signal generator 42. The decoding matrix is used to generate an adjustable color signal C' which is expressed by

C' = kC + (1-k)Y .... (1)

from the color signals C supplied from the frame memories (image memories) 12R, 12G, and 12B, respectively. In expression (1), k is a coefficient for setting the amount of correction, and has a value in the range of 0 ≤ k ≤ 1.

When the value of k is set to 1, a strong correction is performed, namely, the color signals from the frame memories (image memories) 12R, 12G and 12B are directly output as outputs 43R, 43G and 43B, respectively, from the correction setting device 44.

For k = 0, a weak correction is carried out and the color signals from the frame memories 12R, 12G and 12B, respectively, are replaced by the luminance signal Y from the luminance signal generator 42 in the correction adjuster 44, which thereby produces the outputs 43R, 43G and 43B.

The correction setting device 44 forwards the outputs 43R, 43G and 43B to the cumulative histogram generators 32R, 32G and 32B, respectively. In this example, a controller 82 with the correction setting means 44, the color correctors 16R, 16G and 16B; the gradation correctors 18R, 18G and 18B; and a memory 84 which stores the value k of the decoding matrix used to set the correction in the correction setting means 44 and various parameter values set as the color correction matrix MTXA and the look-up tables LUT-2R, LUT-2G and LUT-2B.

The controller 82 is connected to an input unit 86 such as a keyboard to be used by an operator to, for example, select or change the value k of the decoding matrix and image processing parameters such as the look-up table LUT-2R, and to input a necessary instruction such as an instruction to record an image.

When the operator sets the value of k of the decoding matrix to be stored in the correction setting means 44 to 1 with the input unit 86, the mode of operation is set for the strong correction as described above, and therefore the respective color signals of the frame memories (image memories) 12R, 12G and 12B are directly output as outputs 43R, 43G and 43B from the correction setting means 44. As a result, the cumulative histogram generators 32R, 32G and 32B generate cumulative histograms using the color signals from the frame memories (image memories) 12R, 12G and 12B, respectively. Based on the highlight and shadow points, which are then determined using the cumulative histograms, the lookup table converters 34R, 34G and 34B generate the lookup tables LUTR, LUTG and LUTB, which are forwarded to the gradation correctors 14R, 14G and 14B, respectively.

Consequently, the gradation correctors 14R, 14G and 14B can appropriately correct the gradation of the respective color signals; however, for example, when an image has a bright part of a pure color having a high color saturation (chroma saturation), overcorrection results and the color balance is lost, resulting in a color tone different from that of the actual object.

When the operator inputs a value of 0 for the value of k of the decoding matrix of the correction setting means 44 with the input means, a weak correction is effected as described above to replace the color signals from the frame memories 12R, 12G and 12B, respectively, with the luminance signal Y in the correction setting means 44 so that it is passed to the cumulative histogram generators 32R, 32G and 32B, respectively. Consequently, the cumulative histogram generators 32R, 32G and 32B, respectively, generate cumulative histograms using the luminance signal Y from the luminance signal generator 42 to determine the highlight and shadow points. Based on the highlight and shadow points, the lookup table converters 34R, 34G and 34B generate the lookup tables LUTR, LUTG and LUTB, respectively, to be supplied to the gradation correctors 14R, 14G and 14B, respectively.

As a result, the gradation correctors 14R, 14G, and 14B, respectively, effect the same correction in the gradation of the respective color signals to correct only the brightness, so that the resulting signals are generated without correcting the color balance.

If the value of k is set to a suitable value between 0 and 1, the above disadvantage can be eliminated. That is, in The gradation correctors 14R, 14G and 14B can be used to correct the gradation and color balance of the respective color signals; furthermore, in this case, no overcorrection is caused.

Since the value of k is desirably set to about 0.4 in ordinary cases, the value of k need only be set to about 0.4 if the operator does not adjust the value of k by checking an image on a video monitor 80.

In other cases, the value of k is set as follows.

The operator checks the image on the video monitor 80 and decides whether the displayed image contains a completely white part. If so, the value of k is set to 1 or a value close to 1 with the input unit 86. If the completely white part is included, the operator sets k to 1 so as to directly output the color signals of the frame memories (image memories) 12R, 12G, and 12B, respectively, from the correction setting device 44 to generate cumulative histograms. Based on the highlight and shadow points determined from the cumulative histograms, the look-up tables are generated to enable the gradation correctors 14R, 14G, and 14B to appropriately correct the gradation of the respective color signals.

In a case where the image does not have a completely white part, but contains a grey part similar to white, which is represented by R = G = B = 0.9, for example; and the image also has a red part with a high color saturation represented by R = 1.0 and G = B = 0, if k is set to a value near 1, the above-described overcorrection is performed and an image having a weak red and a strong cyan is produced. In this case, therefore, the value k is set to a small value to avoid the overcorrection. If k is specified as a small value, since the highlight point is detected from a signal near the luminance signal Y to be set to a lower point, it is desirable to set high values for the values D of the output density produced by the gradation correctors 14R, 14G and 14B for the highlight points, respectively.

Furthermore, the value of k may be set in advance to 1 or a value close to 1, so that only when the operator detects that a completely white portion is missing from the image displayed on the video monitor 80, the value of k is changed to a smaller value.

In a case where an image as described above contains a slightly gray, almost white portion represented by R = G = B = 0.9 and a red portion with a high color saturation represented by R = 1.0 and G = B = 0, the correction setting means 44, when k is set to 0.4, for example, produces the output signals as follows.

For the luminance signal Y of the part with R = G = B = 0.9 the following applies.

R = G = B = 0.9 is substituted into Y = 0.3R + 0.59G + 0.11B to obtain Y = 0.9. As a result, the Correction setting device 44 a set color signal (chroma signal)

C' = kC + (1-k)Y

which has an R component R' as follows.

= 0.4 x 0.9 + (1-0.4) x 0.9 = 0.9.

On the other hand, the luminance signal Y of the part associated with the expression G = B = 0 is calculated by substituting R = 1.0 and G = B = 0 in the expression Y = 0.3R + 0.59G + 0.11B to obtain Y = 0.3. As a result, the correction adjusting means 44 generates an adjusted color signal

C' = kC + (1-k)Y

the following R has component R'.

R' = 0.4 x 1 + (1-0.4) x 0.3 = 0.58

As a result, for the adjusted color signal R' supplied from the correction adjuster 44, the part where R = G = B = 0.9 has a higher level than the part where R = 1.0 and G = B = 0, and therefore it is not possible to mistakenly regard the part represented by R = 1.0 and G = B = 0 as a highlight point. Even in a case where an image has a strong red portion with a high color saturation, the correction in the gradation correctors 14R, 14G and 14B does not result in overcorrection, thus preventing loss of color balance.

As described above, by setting k to an appropriate value, the gradation correction necessary for the input video signals due to the different conditions under which an object is shot can be correctly achieved, thus minimizing the dependence on the scene and therefore the shot image thus obtained does not appear unnatural.

In addition, as already described above, when the operator checks an image displayed on the video monitor 80 to adjust the value of k, the amount of color correction can be arbitrarily set to achieve the optimum gradation correction according to the image.

Although the embodiments have been described with reference to a method and apparatus used in an image recording system, the method and apparatus for processing an image according to the present invention is not limited to a system for recording a visual image on an image recording medium, namely, the method and apparatus can be used to process various images displayed on a cathode ray tube.

Claims (28)

1. An image processing method for effecting image processing of input video signals (11R, 11G, 11B) based on look-up table data to produce output signals, the image processing method being used in an image recording method that reproduces a visualized image on an image recording medium (30), the method comprising the steps of: Receiving input video signals (11R, 11G, 11B) to determine highlight points and shadow points of the input video signals (11R, 11G, 11B), wherein the determination of the highlight points and the shadow points is based on histograms generated with the input video signals (11R, 11G, 11B); converting standard lookup table data into lookup table data associated with the input video signals (11R, 11G, 11B) based on the highlight points and the shadow points; performing image processing of the input video signals (11R, 11G, 11B) on the basis of look-up table data associated with the input video signals (11R, 11G, 11B), wherein the image processing, based on the look-up table data, converts the input video signals (11R, 11G, 11B) into signals using the look-up table data, representing an intensity of a light to be irradiated onto the image recording medium (30), and achieving a gradation correction of the input video signals (11R, 11G, 11B) depending on a recording condition to generate a density signal, the method being characterized in that the step of determining the highlight points and the shadow points is performed on the basis of cumulative histograms generated with the input video signals, at least one of the highlight point and the shadow point being determined in dependence on at least two points on the cumulative histogram, these two points being in the vicinity of the highlight point or the shadow point to be determined; that the conversion of the standard lookup table is carried out by converting a standard lookup table L0 using a function transformation which is given by the expression L (v) = S (c L0 (a (v + b)) + d) to obtain look-up table data L(v) associated with the input video signals (11R, 11G, 11B), wherein the parameters a, b, c and d are determined by passing the function represented by the expression through the highlight points and the shadow points, and wherein the standard look-up table data is a video gamma curve represented by an expression D = -2.2 log v is pictured. 2. A method according to claim 1, characterized in that in the step of determining the highlight points and the shadow points, the highlight point CH is compared with an expression CH = C2(C2-C3)/(C1-C3) + C1(C1-C2)/(C1-C3) where C1, C2 and C3 represent input video signal levels at points of frequency 99%, 95% and 90% in the cumulative histograms, respectively. 3. A method according to claim 2, characterized in that the values of (C2-C3)/(C1-C3) and (C1-C2)/(C1-C3) are each 1/2 and the highlight point CH is expressed as CH = 1/2(C2+C1) is calculated. 4. A method according to at least one of claims 1 to 3, characterized in that the method further comprises the following step: an input signal level change buffer processing for minimizing a change in a signal level of the input video signal (11R, 11G, 11B), wherein the step of determining the highlight points and the shadow points is carried out on the basis of the input video signals (39R, 39G, 39B) for which the change in signal level is reduced in the step of the input signal level change buffer processing. 5. A method according to claim 4, characterized in that the step of input signal level change buffer processing uses sharpness circuits (40R, 40G, 40B) to change the sharpness coefficients set for the sharpness circuits (40R, 40G, 40B), thereby smoothing the change in the signal level of the input video signals (11R, 11G, 11B). 6. A method according to claim 4 or 5, characterized in that the step of input signal level change buffer processing is effected by the use of a low pass filter. 7. A method according to at least one of claims 4 to 6, characterized in that the step of input signal level change buffer processing reduces the change in signal level only for a luminance signal (41) included in the input video signals (11R, 11G, 11b). 8. A method according to at least one of claims 1 to 7, characterized in that the method further comprises the following step: Receiving the input video signals (11R, 11G, 11B) to effect color correction level adjustment of the input video signals (11R, 11G, 11B), thereby generating output signals (43R, 43G, 43B), wherein the step of determining the highlight points and the shadow points is carried out on the basis of the signals (43R, 43G, 43B) for which the color correction level adjustment is effected in the step of generating the color correction level adjusted signals. 9. A method according to claim 8, characterized in that the step of generating the color correction level adjusted Signals For each chroma signal of the input video signals (11R, 11G, 11B), taking into account a value of a signal representing a brightness of the input video signals (11R, 11G, 11B), the color correction level adjustment can be set to a desired level. 10. A method according to claim 9, characterized in that the step of generating the color correction level adjusted signal is performed at each chroma signal C of the input video signals (11R, 11G, 11B) in consideration of a luminance signal Y of the input video signals (11R, 11G, 11B) to achieve the color correction level adjustment according to an expression C' = kC + (1-k)Y, wherein the value of k is changed to control the degree of the color correction level adjustment. 11. A method according to claim 10, characterized in that the step of generating the color correction level adjusted signal enables an operator to adjust the value for k by checking a corrected image. 12. A method according to claim 11, characterized in that the value of k is set to 1 in advance so that when the operator confirms that no white highlight exists in the corrected image, the value of k is changed to a small value. 13. A method according to claim 10, characterized in that the value for k is set in advance to approximately a middle point between 1 and 0. 4. A method according to claim 9, characterized in that the step of generating the color correction level adjusted signal is performed at each chroma signal of R, G and B of the input video signals (11R, 11G, 11B) in consideration of a value of a signal 1/3 (R+G+B) representing a brightness of the input video signals (11R, 11G, 11B), wherein the color correction level adjustment can be set to a desired level. 15. Image processing device with: Image processing means (14R, 14G, 14B) for effecting image processing of input video signals (11R, 11G, 11B) based on look-up table data to generate output signals, the image processing means (14R, 14G, 14B) being used in an image recording system that reproduces a visualized image on an image recording medium (30), and the image processing means (14R, 14G, 14B) comprising gradation correction means storing look-up table data to convert input video signals into a signal representing an intensity of light to be irradiated onto the image recording medium using the gradation correction means and performing gradation correction of the input video signals in accordance with a recording condition, thereby generating a density signal; Highlight point and shadow point detecting means (32R, 32G, 32B) for receiving the input video signals (11R, 11G, 11B) to set highlight points and shadow points for each of the input video signals (11R, 11G, 11B), the highlight point and shadow point detecting means (32R, 32G, 32B) histogram generating means (32R, 32G, 32B) for generating histograms from the input video signals (11R, 11G, 11B) to determine highlight points and shadow points based on the histograms generated by the histogram generating means (32R, 32G, 32B); Standard lookup table storage means (36) for storing standard lookup table data; Look-up table converting means (34R, 34G, 34B) for converting standard look-up table data, based on the highlight points and the shadow points that have been determined, into look-up table data associated with the input video signals (11R, 11G, 11B) to supply the obtained look-up table data to the image processing means (14R, 14G, 14B), the image processing apparatus being characterized in that the histogram generating means comprise cumulative histogram generating means (32R, 32G, 32B) which generate cumulative histograms with the input video signals (11R, 11G, 11B) in order to determine the highlight points and the shadow points on the basis of the cumulative histograms; that the highlight point and the shadow point determining means (32R, 32G, 32B) determine at least one of the highlight point and the shadow point based on at least two points on the cumulative histogram, the at least two points being in the vicinity of the highlight point or shadow point to be determined; and, that the lookup table conversion means (34R, 34G, 34B) the standard lookup table data L0 with a function transformation which is defined by an expression L (v) = S ( c L0 (a (v + b)) + d) to obtain look-up table data L (v) associated with the input video signals (11R, 11G, 11B), wherein the parameters a, b, c and d are determined by passing the function represented by the expression through the highlight points and the shadow points, and wherein the standard look-up table data is a video gamma curve represented by an expression D = -2.2 log v is pictured. 16. A device according to claim 15, characterized in that the highlight point and shadow point determining means (32R, 32G, 32B) compares the highlight point CH with an expression CH = C2(C2-C3)/(C1-C3) + C1(C1-C2)/(C1-C3) where C1, C2 and C3 represent input video signal levels at points of frequencies 99%, 95% and 90% in the cumulative histograms, respectively. 17. A device according to claim 16, characterized in that (C2-C3)/(C1-C3) and (C1-C2)/(C1-C3) are each 1/2 and the highlight CH is calculated with an expression CH = 1/2(C2+C1). 18. An apparatus according to at least one of claims 15 to 17, characterized in that the apparatus further comprises input signal level change buffer processing means (40R, 40G, 40B) for minimizing a change in a signal level of the input video signals (11R, 11G, 11B), wherein the highlight point and shadow point determining means (32R, 32G, 32B) set the highlight points and the shadow points on the basis of the input video signals (39R, 39G, 39B) for which the change in signal level has been reduced by the input signal level change buffer processing means (40R, 40G, 40B). 19. An apparatus according to claim 18, characterized in that the input signal level change buffer processing means (40R, 40G, 40B) comprise sharpening circuits (40R, 40G, 40B) loaded with variable sharpening coefficients, the sharpening circuit being capable of smoothing the change in the signal level of the input video signals (11R, 11G, 11B). 20. An apparatus according to claim 18 or 19, characterized in that the input signal level change buffer processing means (40R, 40G, 40B) are low-pass filters. 21. An apparatus according to at least one of claims 18 to 20, characterized in that said input signal level change buffer processing means (40R, 40G, 40B) effect processing to reduce the change in signal level only for a luminance signal included in the input video signals (11R, 11G, 11B) input to said cumulative histogram generating means (32R, 32G, 32B). 22. An apparatus according to at least one of claims 15 to 21, characterized in that the apparatus further comprises color correction level adjusting means (44) for receiving the input video signals (11R, 11G, 11B) to effect color correction level adjustment of the input video signals (11R, 11G, 11B), thereby producing output signals (43R, 43G, 43B), wherein the highlight point and shadow point determining means (32R, 32G, 32B) set the highlight points and the shadow points on the basis of the signals (43R, 43G, 43B) for which the color correction level adjustment has been effected in the color correction level adjusting means (44). 23. An apparatus according to claim 22, characterized in that the color correction level adjusting means (44) effects the color correction level adjustment for each chroma signal of the input video signals (11R, 11G, 11B) in consideration of a value of a signal representing a brightness of the input video signals (11R, 11G, 11B), whereby the color correction level adjustment can be set to a desired level. 24. An apparatus according to claim 23, characterized in that the color correction level adjusting means (44) effects the color correction level adjustment on each chroma signal C of the input video signals (11R, 11G, 11B) in consideration of a luminance signal Y of the input video signals (11R, 11G, 11B) according to an expression C' = kC + (1-k)Y, the value of k being changed to control the degree of the color correction level adjustment. 25. An apparatus according to claim 24, characterized in that the color correction level setting means (44) enables an operator to set the value of k to be used for the color correction level setting by checking an image corrected by the gradation correction means (14R, 14G, 14B). 26. An apparatus according to claim 25, characterized in that the value of k is set to 1 in advance so that when the operator confirms that no white highlight exists in the image corrected by the gradation correction means (14R, 14G, 14B), the value of k is changed to a small value. 27. A device according to claim 24, characterized in that the value of k is set approximately at a midpoint between 1 and 0. 28. An apparatus according to claim 23, characterized in that the color correction level adjusting means (44) carries out the color correction level adjustment on each of the chroma signals R, G and B of the input video signals (11R, 11G, 11B) in consideration of a value of a signal 1/3 (R+G+B) representing a brightness of the input video signals, whereby the color correction level adjustment can be set to a desired level.